I had to shutdown my server in 2021 and have just recently revived it. In the process of reviving it I was on version 6.12.5 and have slowly stepped it up to the current version 7.0.1. No issues in the upgrade process.

I turned on the array on maintenance mode and have ran it completely with write/fix mode on and returned with 355466933 errors. I then ran the array with the write/fix mode off with over 100 million errors. 

I ran the xfs_repair on all 3 drives and 2 of them had errors. Zeroing out the log and repairing them "fixed" the drives. 

I suspect that my drives are failing, I have 3 - 8tb wd red hard drives and 1 parity. I have purchased 2 new drives to replace them but I want to do so in stages so I dont lose the data I have. Im wondering if the parity drive is bad?

I tried to run the parity check when the array was online and it came back with a bunch of errors like:

"Mar  2 23:32:20 Tower kernel: md: disk0 read error, sector=681502728"

I have checked the SATA cables and confirmed they are firmly in the socket.

I reran the xfs_repair and it failed, performed the repairs, and it is now clean.

 

I am looking for some guidance as I get reacquainted with running unraid.

I have attached my diagnostics that hopefully can illuminate the path forward.

tower-diagnostics-20250302-2145.zip

There are multiple issues with parity and disk3, check/replace cables, both power and SATA and post new diags after array start.

Start the array in maintenance mode and collect diags, or start the array not in maintenance mode?

Normal mode.

Here are the new diags with the array started not in maintenance mode.

tower-diagnostics-20250305-1200.zip

No issues so far, SMART looks good for all disks, do try re-syncing parity, any errors post new diags before rebooting.

We are 3 min into the read check and it has over 2 million errors on disks. Want me to wait to finish the check or cancel and post diags?

Edited March 6, 20251 yr by frozenfury

If there are read or write errors, post them now.

Here you go. It had 1 hour and 20 min left to finish so the scan was aborted.

tower-diagnostics-20250306-0054.zip

There are issues with multiple disks on the Marvell controller, they are known to be problematic, try connecting tose disks to the Intel SATA controllers instead.

It didnt have any errors this time. Switching to the blue Intel SATA controllers looks to have solved the issue, but Ive included the diags again. After it completed the read checks with no errors the Parity device still shows as disabled.

tower-diagnostics-20250307-0900.zip

Edited March 7, 20251 yr by frozenfury

Looks good so far, regarding parity it must be re-enabled:

 

https://docs.unraid.net/unraid-os/manual/storage-management#rebuilding-a-drive-onto-itself

 

 

After rebuilding the parity disk its messed up my docker service. I have created new directories based on the solution I found here: https://forums.unraid.net/topic/133155-can-not-activate-dockervm-path-does-not-exist/ That has seemed to resolve the visible error. However when I attempt to activate the docker service it fails to start but there is nothing in the syslogs. I have also removed the docker.img at the end and deleted the vDisk file. It doesnt look like it generated a docker.img file in the location specified on the settings.

 

Mar 12 02:36:00 Tower emhttpd: shcmd (576): /etc/rc.d/rc.avahidaemon reload
Mar 12 02:36:00 Tower unassigned.devices: Updating share settings...
Mar 12 02:36:00 Tower unassigned.devices: Share settings updated.
Mar 12 02:36:06 Tower ool www[2690776]: /usr/local/emhttp/plugins/dynamix/scripts/emcmd 'cmdStatus=Apply'
Mar 12 02:36:06 Tower emhttpd: Starting services...
Mar 12 02:36:06 Tower emhttpd: shcmd (583): /etc/rc.d/rc.samba reload
Mar 12 02:36:07 Tower emhttpd: shcmd (587): /etc/rc.d/rc.avahidaemon reload
Mar 12 02:36:07 Tower unassigned.devices: Updating share settings...
Mar 12 02:36:07 Tower unassigned.devices: Share settings updated.

 

tower-diagnostics-20250312-0237.zip

Edited March 12, 20251 yr by frozenfury

Docker service was running before, reboot and post new diags after array start